[PATCH 11/34]bnx2x: Endianness issues

[PATCH 11/34]bnx2x: Endianness issues

[Eilon Greenstein]

Adding missing le_to_cpu and disabling wrong HW endianity flag (the two complete
each other)

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/bnx2x_main.c |   11 ++++++-----
 1 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/net/bnx2x_main.c b/drivers/net/bnx2x_main.c
index ba56d1c..d32b98f 100644
--- a/drivers/net/bnx2x_main.c
+++ b/drivers/net/bnx2x_main.c
@@ -1454,7 +1454,7 @@ static int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
 		DP(NETIF_MSG_RX_STATUS, "CQE type %x  err %x  status %x"
 		   "  queue %x  vlan %x  len %u\n", CQE_TYPE(cqe_fp_flags),
 		   cqe_fp_flags, cqe->fast_path_cqe.status_flags,
-		   cqe->fast_path_cqe.rss_hash_result,
+		   le32_to_cpu(cqe->fast_path_cqe.rss_hash_result),
 		   le16_to_cpu(cqe->fast_path_cqe.vlan_tag),
 		   le16_to_cpu(cqe->fast_path_cqe.pkt_len));
 
@@ -2837,8 +2837,10 @@ static void bnx2x_attn_int_deasserted(struct bnx2x *bp, u32 deasserted)
 static void bnx2x_attn_int(struct bnx2x *bp)
 {
 	/* read local copy of bits */
-	u32 attn_bits = bp->def_status_blk->atten_status_block.attn_bits;
-	u32 attn_ack = bp->def_status_blk->atten_status_block.attn_bits_ack;
+	u32 attn_bits = le32_to_cpu(bp->def_status_blk->atten_status_block.
+								attn_bits);
+	u32 attn_ack = le32_to_cpu(bp->def_status_blk->atten_status_block.
+								attn_bits_ack);
 	u32 attn_state = bp->attn_state;
 
 	/* look for changed bits */
@@ -2886,7 +2888,7 @@ static void bnx2x_sp_task(struct work_struct *work)
 	if (status & 0x2)
 		bp->stats_pending = 0;
 
-	bnx2x_ack_sb(bp, DEF_SB_ID, ATTENTION_ID, bp->def_att_idx,
+	bnx2x_ack_sb(bp, DEF_SB_ID, ATTENTION_ID, le16_to_cpu(bp->def_att_idx),
 		     IGU_INT_NOP, 1);
 	bnx2x_ack_sb(bp, DEF_SB_ID, USTORM_ID, le16_to_cpu(bp->def_u_idx),
 		     IGU_INT_NOP, 1);
@@ -5177,7 +5179,6 @@ static int bnx2x_init_common(struct bnx2x *bp)
 	REG_WR(bp, PXP2_REG_RQ_SRC_ENDIAN_M, 1);
 	REG_WR(bp, PXP2_REG_RQ_CDU_ENDIAN_M, 1);
 	REG_WR(bp, PXP2_REG_RQ_DBG_ENDIAN_M, 1);
-	REG_WR(bp, PXP2_REG_RQ_HC_ENDIAN_M, 1);
 
 /*	REG_WR(bp, PXP2_REG_RD_PBF_SWAP_MODE, 1); */
 	REG_WR(bp, PXP2_REG_RD_QM_SWAP_MODE, 1);
-- 
1.5.4.3

Re: [PATCH 11/34]bnx2x: Endianness issues

[David Miller]

From: "Eilon Greenstein" <eilong@broadcom.com>
Date: Wed, 14 Jan 2009 18:43:15 +0200

> Adding missing le_to_cpu and disabling wrong HW endianity flag (the
> two complete each other)
>
> Signed-off-by: Eilon Greenstein <eilong@broadcom.com>

Applied.

Re: [PATCH 11/34]bnx2x: Endianness issues

[Christoph Hellwig]

On Wed, Jan 14, 2009 at 06:43:15PM +0200, Eilon Greenstein wrote:
> Adding missing le_to_cpu and disabling wrong HW endianity flag (the two complete
> each other)

Did you run sparse over the driver to verify you've sorted out all
endianess issues?

Re: [PATCH 11/34]bnx2x: Endianness issues

[Eilon Greenstein]

On Sun, 2009-01-18 at 08:59 -0800, Christoph Hellwig wrote:
> On Wed, Jan 14, 2009 at 06:43:15PM +0200, Eilon Greenstein wrote:
> > Adding missing le_to_cpu and disabling wrong HW endianity flag (the two complete
> > each other)
> 
> Did you run sparse over the driver to verify you've sorted out all
> endianess issues?
> 
No - those are errors with the FW/HW and not general issues. When
running sparse with CF=-D__CHECK_ENDIAN__ I get "too many warnings"...
So I cannot send the patch right now. I will definitely add it to my
queue and I hope to sort it out soon.

Eilon

Re: [PATCH 11/34]bnx2x: Endianness issues

[Harvey Harrison]

On Mon, 2009-01-19 at 12:13 +0200, Eilon Greenstein wrote:
> On Sun, 2009-01-18 at 08:59 -0800, Christoph Hellwig wrote:
> > On Wed, Jan 14, 2009 at 06:43:15PM +0200, Eilon Greenstein wrote:
> > > Adding missing le_to_cpu and disabling wrong HW endianity flag (the two complete
> > > each other)
> > 
> > Did you run sparse over the driver to verify you've sorted out all
> > endianess issues?
> > 
> No - those are errors with the FW/HW and not general issues. When
> running sparse with CF=-D__CHECK_ENDIAN__ I get "too many warnings"...
> So I cannot send the patch right now. I will definitely add it to my
> queue and I hope to sort it out soon.
> 
> Eilon

I'll take a look as well.  BTW, make sure you don't have the trace all
if statements tracer enabled in your config or you'll be drowned in
shadowed variable warnings and won't be able to see anything from your
driver.

Harvey